Introduction

The present study explores the relationship between sleep quality and memory performance in children with autism and within the broader autism phenotype. Previous research has consistently demonstrated that sleep disturbances are prevalent in autistic populations (Richdale & Schreck, 2009; Cohen et al., 2014). However, only a limited number of studies have directly investigated whether poor sleep is associated with memory deficits, which are a prominent concern for this population. Given the significant heterogeneity in memory functioning among autistic individuals, it has been proposed that sleep may play a crucial role in explaining this variability. A more nuanced understanding of the relationship between sleep and memory could enhance the well-being of autistic individuals and contribute to broader theories of cognition and behavior in clinical populations.

To address these aims, this dissertation analyzes quantitatively collected data from a cohort of children, with and without autism, using objective sleep measures (e.g., sleep efficiency, fragmentation index, actual sleep time, and sleep latency), cognitive assessments (WASI Block Design and Vocabulary subtests), autism traits (Autism Spectrum Quotient, AQ), and a range of memory-related outcomes (AMAP subscales). By examining these associations, the study aims to determine whether variability in sleep quality explains individual differences in memory performance and whether autism trait level moderates these effects.

Hypotheses

• Hypothesis 1 (Main Effect - Sleep and Memory): There will be a significant positive correlation between objective measures of sleep quality (e.g., sleep efficiency, total sleep time, fragmentation index) and memory performance (e.g., relational memory, visual recognition, temporal order, spatial memory, scene recognition, associative memory) in the overall sample of children.
• Hypothesis 2 (Group Differences in Memory): Children with autism and children with high autistic traits will exhibit lower scores on certain memory subtests, particularly relational memory, compared to neurotypical controls.
• Hypothesis 3 (Group Differences in Sleep): Children with autism and children with high autistic traits will demonstrate poorer objective sleep quality (e.g., lower sleep efficiency, higher fragmentation index, longer sleep latency) compared to neurotypical controls.
• Hypothesis 4 (Moderation - Autism Traits): The relationship between sleep quality and memory performance will be moderated by autism trait level (as indexed by the AQ questionnaire), with sleep quality having a more pronounced impact on memory deficits in children with higher autistic traits.

Method

Participants

Data for this study were collected from children participating in the “Young Scientist Day” event at Kingston University. The sample included children with an autism diagnosis, children exhibiting high autistic traits but without a formal diagnosis, and neurotypical controls, reflecting the broader autism phenotype. Informed consent was obtained from the parents or legal guardians of all participants in accordance with university ethical standards.

Procedure

Event and Setting

The study took place during the Young Scientist Day, a university-sponsored outreach event designed to engage children in science-based activities. Participants and their families were invited to take part in a series of cognitive and behavioral assessments in a controlled, child-friendly environment at Kingston University.

Sleep Data Collection

• Device: Sleep quality and activity were objectively measured using the MotionWatch 8, a wrist-worn actigraph. Each child wore the MotionWatch on their non-dominant wrist for a continuous period of 7 days.
• Instructions: Families were instructed to keep the device on at all times except during activities that involved full submersion in water (e.g., bathing, swimming) or contact sports. The MotionWatch’s event button was pressed each night at bedtime and again upon waking to accurately mark sleep periods.
• Data Captured: The MotionWatch recorded various sleep parameters including sleep efficiency, total sleep time, sleep latency, and fragmentation index. Compliance and usage guidelines were explained in person, and a brief demonstration was provided to ensure consistency of data collection.

Cognitive and Memory Assessments

• IQ Testing: Each participant completed the Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II). The Block Design and Vocabulary subtests were administered to assess perceptual reasoning and verbal comprehension, respectively.
• Memory Assessment: Memory was assessed using a battery of tasks collectively referred to as the AMAP (Alien Memory Assessment Protocol). These included subtests for relational memory, visual recognition, temporal order, spatial memory, scene recognition, and associative memory.
• Autism Traits: The Autism Spectrum Quotient (AQ) questionnaire was employed to index autistic traits among all children.

Testing Schedule

• On arrival, participants were greeted and provided with an overview of the study.
• The WASI-II and AMAP assessments were conducted in a quiet, distraction-free setting.
• Following cognitive testing, the MotionWatch was fitted, and instructions were provided to the child and parent/caregiver.
• After the 7-day actigraphy monitoring period, MotionWatch devices were returned for data download and analysis.
• All participants were debriefed and thanked for their contribution upon completion.

Data Analysis

Data from the actigraphy devices and cognitive assessments were collated for statistical analysis. Sleep parameters were averaged across the monitoring period. Group differences and associations between sleep quality, memory performance, and autism trait level were examined using appropriate statistical tests, including correlation and regression analyses. This methodology ensured rigorous, objective measurement of sleep patterns alongside robust cognitive and memory assessment in a real-world community setting.

Results

Hypothesis 1: Sleep Quality and Memory Performance (Overall Sample)

Hypothesis 1 posited a significant positive correlation between objective measures of sleep quality and memory performance in the overall sample of children. This hypothesis received partial support.

Pearson's correlation analyses revealed a statistically significant negative correlation between Sleep-Latency_(mins)and Object_in_context_2_Score, r = -0.415, p = .028. This indicates that longer sleep latency (suggesting poorer sleep initiation) was associated with lower performance on this specific memory task. Additionally, a significant negative correlation was observed between Alien_Object_Config_Memory_Average and Object_in_context_1_Score, r = -0.542, p = .003, suggesting an inverse relationship between these two memory domains.

Regarding subjective measures, a significant negative correlation was found between CSHQ_Total (Child Sleep Habits Questionnaire, where higher scores indicate more sleep problems) and OMQ_total (Observed Memory Questionnaire total), r = -0.472, p = .012. This suggests that greater self-reported sleep problems were associated with lower observed memory scores. Furthermore, CSHQ_Total was positively correlated with AQ_total, r = 0.631, p < .001, indicating that children with higher autistic traits tended to report more sleep difficulties.

No other statistically significant positive correlations were found between the remaining objective sleep measures (Sleep_Efficiency, Fragmentaion_Index, Actual_Sleep_time_(Mins)) and the other objective memory measures (Temporal_Order_Score, Visual_Recognition_Score, Spatial_Tot, Scene_Rec_Score, Object_in_context_1_Score) in the hypothesized positive direction.

Hypothesis 2: Group Differences in Memory Performance

Hypothesis 2 predicted that children with autism and high autistic traits (High AQ group) would exhibit lower scores on certain memory subtests, particularly relational memory, compared to neurotypical controls (Low AQ group). This hypothesis received partial support.

Independent samples t-tests revealed no statistically significant differences between the Low AQ and High AQ groups for Alien_Object_Config_Memory_Average (t(26) = 0.540, p = .594, Cohen's d = 0.204) or any other objective memory subtests (Temporal_Order_Score, Visual_Recognition_Score, Spatial_Tot, Scene_Rec_Score, Object_in_context_1_Score, Object_in_context_2_Score).

However, a statistically significant difference was found for OMQ_total, with the High AQ group (M = 79.71, SD = 9.77) reporting significantly lower observed memory scores compared to the Low AQ group (M = 92.43, SD = 8.79), t(26) = 3.261, p = .003, Cohen's d = 1.234. This indicates a large effect size for subjectively observed memory difficulties in the group with higher autistic traits.

Hypothesis 3: Group Differences in Objective Sleep Quality

Hypothesis 3 stated that children with autism and high autistic traits would demonstrate poorer objective sleep quality (e.g., lower sleep efficiency, higher fragmentation index, longer sleep latency) compared to neurotypical controls. This hypothesis received partial support.

Independent samples t-tests on objective sleep measures revealed no statistically significant differences between the Low AQ and High AQ groups for Sleep_Efficiency (t(26) = 0.637, p = .530, Cohen's d = 0.241), Fragmentaion_Index(t(26) = -0.247, p = .807, Cohen's d = -0.093), Actual_Sleep_time_(Mins) (t(26) = 0.732, p = .471, Cohen's d = 0.277), or Sleep-Latency_(mins) (t(26) = -0.545, p = .590, Cohen's d = -0.206).

Conversely, a statistically significant difference was found for CSHQ_Total (Child Sleep Habits Questionnaire), with the High AQ group (M = 59.21, SD = 12.92) reporting significantly more sleep problems compared to the Low AQ group (M= 42.71, SD = 9.28), t(26) = -3.261, p = .003, Cohen's d = -1.234. This indicates a large effect size for subjectively reported sleep problems in the group with higher autistic traits.

Hypothesis 4: Moderation of Sleep-Memory Relationship by Autism Trait Level

Hypothesis 4 proposed that the relationship between sleep quality and memory performance would be moderated by autism trait level, with sleep quality having a more pronounced impact on memory deficits in children with higher autistic traits. This hypothesis received support for the specific relationship examined.

A linear regression analysis predicting Alien_Object_Config_Memory_Average from Sleep_Efficiency, AQ_total, and their interaction was conducted. The overall model was not statistically significant, F(3, 24) = 2.548, p = .080, R² = 0.242. However, the interaction term Sleep_Efficiency ✻ AQ_total was statistically significant, b = -0.023, SE = 0.010, t = -2.282, p = .032. This indicates that the relationship between Sleep_Efficiency and Alien_Object_Config_Memory_Average is significantly moderated by AQ_total. The negative coefficient suggests that as AQ_total increases, the association between Sleep_Efficiency and Alien_Object_Config_Memory_Averageweakens or becomes more negative, implying that the potential benefits of higher sleep efficiency on relational memory are diminished in children with higher autistic traits. Collinearity diagnostics indicated severe multicollinearity (Condition Index = 209.997), a common issue with interaction terms, though the significance of the interaction term remains interpretable.

Discussion

The present study aimed to investigate the complex interplay between sleep quality, memory performance, and autism trait level in a cohort of children, including those with autism and within the broader autism phenotype. The findings provide nuanced insights into these relationships, offering both support for some hypotheses and unexpected results for others, highlighting the heterogeneity within this population.

Contrary to initial expectations, Hypothesis 1, which predicted a broad positive correlation between objective sleep quality and memory performance across the overall sample, received only partial support. While a direct positive link was largely absent for objective measures, a significant negative correlation between Sleep-Latency_(mins) and Object_in_context_2_Score suggests that difficulties in falling asleep may indeed be detrimental to specific memory functions. More notably, the study found a significant negative correlation between subjective sleep problems (higher CSHQ_Total) and observed memory difficulties (lower OMQ_total), aligning with existing literature that often relies on subjective reports of sleep disturbances and their impact on daily functioning. The strong positive correlation between CSHQ_Total and AQ_total further reinforces the notion that children with higher autistic traits tend to experience more perceived sleep problems. This discrepancy between objective and subjective sleep measures is a critical finding, suggesting that while actigraphy may not capture all aspects of sleep disturbance that impact memory, parental reports (CSHQ) and observed memory (OMQ) may be more sensitive indicators of functional impact in this population.

Hypothesis 2, predicting lower memory scores in the high autistic trait group, also received partial support. While objective memory tasks from the AMAP battery did not reveal significant differences between the Low AQ and High AQgroups, the High AQ group did report significantly lower scores on the OMQ_total. This suggests that while children with higher autistic traits may perform comparably on structured, tablet-based memory assessments, their everyday memory functioning, as observed by caregivers, might be more impaired. This finding underscores the importance of considering both objective and subjective measures when assessing cognitive abilities in neurodevelopmental conditions, as laboratory-based tasks may not fully capture real-world challenges.

Similarly, Hypothesis 3, concerning group differences in sleep quality, was partially supported. Objective actigraphy measures of sleep (Sleep_Efficiency, Fragmentaion_Index, Actual_Sleep_time_(Mins), Sleep-Latency_(mins)) did not significantly differentiate the Low AQ and High AQ groups. This could be due to the relatively small sample size, the specific operationalization of "high AQ" versus a formal diagnosis, or the inherent variability in objective sleep parameters even within clinical populations. However, consistent with Hypothesis 2's findings, the High AQ group reported significantly more sleep problems on the CSHQ_Total. This reinforces the idea that subjective experience of sleep disturbance is a prominent feature in children with higher autistic traits, even if objective measures do not always show a clear distinction from neurotypical peers in this sample. This highlights the potential for a disconnect between objective physiological sleep patterns and the lived experience of sleep quality and its impact.

Perhaps the most compelling finding relates to Hypothesis 4, which posited that autism trait level would moderate the relationship between sleep quality and memory performance. This hypothesis was supported for the specific interaction between Sleep_Efficiency and AQ_total in predicting Alien_Object_Config_Memory_Average. The significant negative interaction term suggests that the beneficial effect of higher sleep efficiency on relational memory is attenuated or even reversed as autistic traits increase. This implies that good sleep quality may not confer the same memory advantages to children with higher autistic traits as it might to those with fewer traits. This finding is crucial for understanding the heterogeneity in memory functioning observed in autistic individuals and suggests that sleep may indeed play a unique role in shaping memory outcomes within the broader autism phenotype. It points towards a more complex, moderated relationship rather than a simple direct effect, providing a nuanced understanding of how sleep influences cognition in this population.

Limitations

Despite these valuable insights, the study has several limitations. The sample size of 28 participants, with 14 in each AQ group, is relatively small, which may limit the statistical power to detect smaller effects and generalize findings to a broader population. The reliance on convenience sampling from a "Young Scientist Day" event might introduce selection bias, as participating families may differ systematically from the general population of children with and without autistic traits. While AQ_Group effectively differentiated participants by trait level, the absence of formal diagnoses for all "High AQ" participants means the findings cannot be directly generalized to clinically diagnosed autism spectrum disorder without caution. Furthermore, the objective sleep data, while valuable, may not capture all aspects of sleep quality relevant to memory, and the 7-day actigraphy period might not fully account for long-term sleep variability. The presence of severe multicollinearity in the moderation analysis, though common, also warrants careful interpretation of individual main effects.

Conclusion

In conclusion, this study provides important insights into the relationship between sleep, memory, and autism traits in children. While broad direct associations between objective sleep quality and memory were largely unsupported, subjective reports highlighted significant sleep difficulties and observed memory impairments in children with higher autistic traits. Crucially, the finding that autism trait level moderates the relationship between sleep efficiency and relational memory suggests a more intricate interaction, where good sleep may not uniformly benefit memory across the autism spectrum. These findings underscore the importance of considering both objective and subjective measures of sleep and memory, and suggest that interventions targeting sleep in autistic populations may need to consider individual differences in autism trait severity to optimize cognitive outcomes. Future research with larger, more diverse samples and longitudinal designs is warranted to further elucidate these complex relationships and inform targeted interventions.